The Mixnet v2 has been proposed by Preliminary Research: Mixnet Design Framework - Maximally-decentralized and homogeneous private consensus. Before designing details, all approaches suggested in the proposal should be verified to check if they are actually appropriate to the problems that we aims to solve.

This document describes the proof of concept for the Mixnet Gadget introduced in the proposal. Although this proof of concept focuses on the Mixnet gadget, it is also based on other protocols suggested in the proposal, such as block building and block selection. The proof of concept is going to be done by empirical analysis (i.e. simulations). Although some important components need to be analyzed mathematically, we expect that having a simulation framework is beneficial to find the optimal structures and parameters based on the realistic environments.

Simulation Plan

Goals

Check how well the new design hides who proposed a particular block from network adversaries, and find improvements needed.

The simulation consists of two big parts:

1. Simulating the Mixnet v2 design proposed
2. Simulating adversary models

Related Work

• MiXiM: Mixnet Design Decisions and Empirical Evaluation, 2021
  • https://gitlab.esat.kuleuven.be/Iness.BenGuirat/mixim
  • MiXiM aims to simulate global passive adversaries (GPA) by measuring only the Shannon entropy. For more details of entropy, please see the App. A in the paper.
  • MiXiM is built with SimPy, simulates the network, and logs all information that needs to be analyzed. At the final stage, scripts parse logs to compute the results (entropy).
• Studying the anonymity trilemma with a discrete-event mix network simulator, 2021 (Medium)
  • https://github.com/aniampio/Simulator
  • Almost the same as MiXiM, this simulator focuses on GPA and measures the Shannon entropy. But, older than and seems to have less analysis features than MiXiM.
• ‣, 2023
  • They aim to model and analyze Adversaries with Partial Visibility (APV) based on Syverson’s research “Why I’m not an Entropist”, 2013 that says GPA is unrealistic.
• Waku v2 Adversarial Models and Attack-based Thread List
  • This document categories the adversarial models well and introduces some attacks from the adversaries. Most attacks introduced in the document are not very different from those used in other work above. But, the Neighbourhood Surveillance is interesting and well-explained. The example focuses on the generic p2p communication, but it seems to be useful for our problem as well.
• [ ] Mixnet v2 Design Simulation

To make the simulation simple and flexible, only core components will be implemented. For example, networking between nodes can be simulated as memory operations.

Key Objectives

It is important to check the viability of the new design and determine reasonable design details, especially for the following aspects.

• Overall Objectives
  • Bandwidth efficiency: 250Kbps as a initial target
    • All objectives listed below should be verified in terms of bandwidth efficiency.
  • Attack Prevention (Verifying the security level that our design can provide)
    • It should be evaluated how well the system defends against the attacks listed below.
• Detailed Objectives
  • Encoding a message (a block hash + a proof) along with pre-signed incentivization transactions using the modified version of Sphinx
    • By limiting the size of payload according to the maximum number of mix layers, in order to make all packets have the same size
  • Cover traffic from all nodes pretending to be a block proposer
    • A clear strategy should be determined, especially in terms of timing.
  • Timing obfuscation (Poisson delays)
    • It must be verified whether this is really necessary by measuring its effectiveness.